| Copyright 1997 IEEE. Personal use of this material is permitted.
However, permission to reprint/republish this material for advertising
or promotional purposes or for creating new collective works for resale
or redistribution to servers or lists, or to reuse any copyrighted
component of this work in other works must be obtained from the IEEE. |
 |
 |
This article was published in the June 1997 issue of
IEEE Communications Magazine.
 |
Abstract
Broadband networks are now being used in numerous trials throughout the world to provide interactive video services for residential customers. These trials and small-scale commercial deployments lack novel applications and services that make effective use of network capabilities and attract users to experiment with them. The rapid explosion of the Internet has also played a role in the current lack of interest and investment in broadband video networks. This article examines the lessons learned from the commercial deployment of a switched fiber-coax video dial tone network in Dover Township, New Jersey, USA. Future broadband networks will have to combine the capabilities of video dial tone networks and the Internet at an attractive price. Innovative end-to-end applications that go beyond cable TV and movies on demand are needed to renew commercial interest in residential broadband networks.
Interactive Digital Video Networks: Lessons from a Commercial Deployment
Kamlesh Rath, Don H. Wanigasekara-Mohotti, Roli G. Wendorf, and Dinesh C. Vermaz
Philips Research
During the first half of this decade, the telecommunications industry was very enthusiastic about residential broadband networks. Telecommunications companies wanted to enter the profitable cable TV business. Digital video compression technology and packet-switched network technology had been standardized and were available. It was possible to provide broadband networks to carry switched digital video to residential customers [1, 2]. All these factors together created a frenzy of trials of residential broadband networks in various locations around the world.
These networks were conceived primarily for cable TV applications with some limited interactive capabilities [3, 4]. The level of interactivity was designed to facilitate movies-on-demand applications, and required a huge, expensive infrastructure to support on-demand point-to-point interconnections. Hence, the residential broadband network trials were not successful in unleashing major new markets.
In the meantime, the Internet [5] and the World Wide Web have been expanding rapidly. The popularity of the Web is due to the wide variety of applications and services available through a simple uniform user interface. If residential broadband networks are to be popular, they, too, need to support access to the Internet, and other similar end-to-end applications that are easy to create and use.
This article examines residential broadband networks from a user, application, and service point of view. The second section describes various broadband network architectures for interactive video services. The third section describes the switched fiber video dial tone network deployed in Dover Township, New Jersey. The fourth section describes end-to-end applications that can make effective use of the resources of a broadband network. The fifth section examines down-loaded applications. The sixth section looks at the convergence of the Internet and broadband networks. Finally, the last section explores the future of broadband networks and the applications and services for such networks.
Broadband Network Architectures
Several key aspects of a digital video network are dependent on the architecture used for interconnecting the end user of the services to the service provider. Different kinds of network architectures offer varying amounts of capacity from the consumer to the service provider, from the service provider to the consumer, varying security features, and varying quality of service. These can contribute significantly to the cost of commercial deployment.
For satisfactory delivery of continuous media services like video, the network must provide sufficient capacity from the service provider to the consumer. The network terminal must send information back to the service provider, for example, to request a particular movie or to request download of a particular game. This "back-channel" capacity needed for control is usually much smaller than the forward channel. Video applications that utilize the back channel effectively can be imagined. However, insufficient attention paid to this factor during network design severely restricts successful use of the back channel.
In order to ensure a successful commercial deployment, one needs to consider the privacy and security issues that may be associated with a video network. While these issues are not fundamentally different from those of a traditional data network, the network must provide security mechanisms that prevent neighbors from knowing what an individual is watching, and for the service provider to authenticate that a request indeed originated from (and will be paid for by) the user receiving that service. For home shopping and electronic banking applications, protecting sensitive data such as credit card information is essential.
The management of the network used for video delivery is also an important and essential part of a commercial video network. However, network management of a commercial digital video network is no different from that of a traditional broadband network and will not be examined further in this article.
The different networking architectures that can be used in a digital video network are described in the following subsections.
Switched Fiber Networks
From a pure capacity point of view, a switched optical fiber network from the service provider to the home is the best solution. A switched optical fiber network based on asynchronous transfer mode (ATM) technology is capable of providing multiple connections of variable bandwidth between any pair of points, and offers maximum flexibility for deploying and developing video (and other continuous-media) applications that can be brought to the end consumer. Such a network can carry up to 1 Gb/s in either direction. However, the cost of deploying such a fiber-to-the-home network is very high.
There are two cost-effective solutions for switched fiber networks. The system in the Dover Township uses a fiber-to-the-curb (FTTC) network. Optical fiber runs to curbside units, which transfer the signals onto coaxial cable. The coaxial cable then carries data the last few hundred feet to the subscriber's home. Such a network limits the end user's data delivery rate to the bandwidth of the coaxial cable, which rates in the hundreds of megabits per second.
The other solution terminates the fiber at a local phone exchange, and carries the data using digital subscriber line (xDSL) technologies over twisted copper wire to subscriber homes. These technologies work by placing special equipment at the two ends of the traditional phone company's copper loop to increase the effective throughput. The technologies proposed for the copper loop include ADSL (asymmetric DSL), which can deliver 1.5–9 Mb/s to the home and 16–640 kb/s from the home to the service provider; SDSL (single service DSL) that can deliver up to 2Mb/s in either direction; and very high-speed DSL (VDSL), which can deliver 13–52 Mb/s to the home, and 1.5-2Mb/s from the home to the subscriber.
The xDSL technologies are very attractive from a local telephone company perspective, since they permit the reuse of a very extensive copper-based network. To the cable companies, however, which constitute another significant force in the area of video networking, cable channels offer a preferred alternate architecture, as described in the next section.
Broadcast Networks
There are several broadcast networks currently in existence which are used to deliver audio and video to the home. The two most common networks are the local cable TV networks and the broadcast television networks. The broadcast television network delivers analog signals to the home, but with the advent of high-definition TV (HDTV), digital video over the air is expected to begin before the turn of the century. The cable TV operators already have an established network in place which can broadcast analog video to the home.
Using a cable TV modem, such a network can have a back-channel with a capacity of 200 kb/s–2 Mb/s for the home unit to communicate with the video information server. Available bandwidth from the information provider to the home is in the range of 3–10 Mb/s. Using modulation techniques such as 64QAM (quadrature amplitude modulation), the forward capacity can be as high as 36 Mb/s, and the back-channel capacity can go up to 10 Mb/s. There are two main reasons for the lower speed on the back-channel: first, the back-channel usually uses a frequency range that is more noisy (64QAM cannot be used on the back channel); second, the units in different homes have to compete for access on the back channel. This contention leads to a lower effective throughput per home.
Another class of broadcast networks often used for transmission of digital video is satellite networks, such as DSS (digital satellite system). A competitor to the cable television operators, satellite systems operate by relaying digital programs from a geostationary satellite. The home units have an antenna to receive the satellite signals, and require a special subscription number in order to descramble and view the transmitted program. As in the case of broadcast television networks, there is no back channel available. Satellite systems can transmit as many as 300 television channels to users.
The back-channel problem in satellite systems can be solved by having the network terminal communicate with the service provider over plain old telephone service (POTS) lines using a modem. Using current technology, the analog telephone channel can provide a bandwidth of 33.6 kb/s in the reverse direction. The network terminal unit has to be configured with the phone number of the service provider in order to place the telephone call.
Another technology used for transmission of digital video on the air is multichannel multipoint distribution system (MMDS) or wireless cable. MMDS networks operate in a manner similar to that of a satellite network, except that broadcast channels use microwave frequencies for transmission. A strong transmitter at a central site is capable of carrying a limited number (about 30) of television channels to consumers.
The main limitation of satellite or MMDS systems is that multimedia applications need to be developed to take into account the simplex nature of transmission. Cable network operators using cable modems have more flexibility in terms of the services they can provide to the end customer.
Deployment at Dover Township, New Jersey
A residential video broadband network has been deployed by Bell Atlantic in Dover Township, New Jersey, starting in January 1996. This network supports an enhanced cable TV service, including digital TV (up to 384 channels) and interactive capabilities. The Bell Atlantic deployment competes against an existing analog cable system. As of July 1996, approximately 4500 customers had subscribed to the service, and the number is expected to rise to 20,000 over the next two years or so. Philips Electronics has supplied the TV set-top boxes (also called digital entertainment terminal or network terminal) for this deployment.
The overall end-to-end network is shown in Fig. 1. The unit residing in the customer's home is the network terminal, which is connected to a standard analog TV. The user interacts with the network terminal through a remote control unit, or via keys on the network terminal. A part of the network terminal is a replaceable communication module, which corresponds to the networking environment.
The communication module connects the network terminal to a central office switch through an FTTC network which terminates in a curbside optical network unit, and is connected from the curb to the communication module by coax cable. All communications are through a one-way broadband video channel (solid arrows in Fig. 1) and a low-bandwidth two-way signaling channel (dashed arrows in Fig. 1), using a vendor-proprietary protocol.
The central office switch acts as an access concentrator and a switching system for the network terminals in a particular geographical area, as in the case of telephony networks. It allows the customer to access one of a number of services by selectively sending the requested channels to the network terminal.
The central office switch is connected to the various service providers through Bell Atlantic's video dial tone (VDT) network. The VDT network consists of a switched broadband, fiber video network for carrying high-bandwidth, unidirectional video information (solid arrows in Fig. 1), and a packet-switched signaling network for bidirectional, symmetrical transmission of information (dashed arrows in Fig. 1). This VDT network is capable of a 180 Mb/s downstream channel and a 10 kb/s upstream signaling channel to/from the set-top box. Part of the downstream channel capacity (135 Mb/s) is allocated to three video channels, and the remaining capacity (45 Mb/s) is allocated to the downstream signaling channel. The fiber optic network is terminated on the curbside by an optical network unit that can support eight such connections, or a total of 24 set-top boxes per optical network unit. The headend equipment stationed in a neighborhood can support 32 optical network units. The signaling network is a low-speed (16 kb/s) bidirectional X.25 network [6]; for more details, see [7].
The VDT network supports two main classes of services: broadcast services, such as broadcast video and broadcast text; and point-to-point on-demand services, such as interactive text and video on demand. Broadcast video services currently consist of cable TV and distribution of new releases of network terminal operating system software. Broadcast text and interactive text services only use the signaling network. All services are discussed in more detail later in the article.
The networking infrastructure required to support point-to-point, on-demand services is more complex, because the video and signaling channel connections have to be dynamically set up and torn down. These services are not yet available in the Dover Township deployment because of lack of consumer interest.
To support on-demand services, Bell Atlantic provides a level 1 gateway (L1GW) [7] connected to the VDT signaling network, which interacts with network terminals and allows them to access all available on-demand services supplied by video information providers (VIPs). The on-demand broadband channels are shown as dashed arrows in Fig. 1. The L1GW allows menuing and navigation of available VIP services, service selection, network terminal authorization, and so on. It also interacts with the video network to allow establishment and teardown of a video connection between a VIP and a network terminal.
VIP servers are connected to the VDT network through level 2 gateways (L2GWs). An L2GW provides access to information from one or more VIP servers. It allows menuing and navigation of available VIP selections (e.g., available videos), service selection, network terminal authorization, and so on. The L2GW interacts with the L1GW for network control.
Applications
From a consumer's perspective, the most important aspect of an interactive video network is its applications. The underlying network and video compression technologies for interactive video networks have been around for a few years. The consumer must now see applications that use the broadband capacity effectively and deliver information and programming that is substantially better than is available today on any medium.
Cable TV and More
Initial deployments of interactive video networks, including the Dover Township deployment, were primarily designed to provide cable TV service to consumers at home. The telecommunication companies were looking to take business away from cable operators by providing a large number of channels of better-quality digital programs. This required an enormous investment in fiber networks and switches for providing cable-TV service. The investment has now shifted to satellite networks because they are very cost effective and easy to deploy, and provide the same quality of cable TV service.
The Dover service has 384 cable TV channels divided among different VIPs. A channel change request from the user is passed on to the central office equipment, which switches the requested video channel to the home terminal.
The Dover network also offers a simple interactive text facility along with the cable TV service. This facility can be used to make point-to-point connections over the X.25 network between a consumer and an interactive text provider by choosing a menu option from a channel. The network terminal behaves like a small VT100 terminal. Simple text-based services can be provided using this facility to order pizzas, movie tickets, and so on. This form of interactivity is very primitive compared to the Internet. Other interactive applications that are rich in video, graphics, and text could be developed to make better use of the capability of broadband networks. Note that from a technical point of view, it is feasible to provide Internet access and other more interesting applications on this network.
Video on Demand
During the deployment of interactive video networks it was expected that video on demand would be the "killer application" which would generate profits for the industry. It was later realized that true video on demand was too expensive, and consumers were not willing to pay the extra price for ordering a movie and watching it instantly. The expense was due to the requirement for a high-bandwidth point-to-point connection between a server and a consumer.
As a result, the concept of near video on demand was born, which meant that popular movies are distributed on many channels with staggered start times. A consumer can order a movie and wait a few minutes for the next start. This concept does not require point-to-point connections and consequently is cheaper than video on demand. Again, this service does not offer anything novel to consumers because similar services are available on cable systems today.
Home Shopping
Consumers spend a lot of money today on shopping from home over the cable networks. The number of cable channels dedicated to home shopping is increasing. These channels broadcast product information and show video images of the products. The accompanying commentary encourages people to call a telephone number to purchase the product.
This form of home shopping has very limited interactivity. Every consumer is shown the same broadcast material, and is not able to choose products to preview. Also, a consumer has to use the telephone for more information or to make purchases. Interactive video networks can provide much more interactivity, enabling consumers to choose items to preview, and place orders on the home terminal. This requires a point-to-point broadband channel from a server to a home terminal (described in the next subsection). This also requires a hyper-linked navigation environment that allows consumers to choose options easily and browse through available product catalogs. Such applications can be very popular and can pay for the investments required to build large broadband networks.
Electronic Banking
Currently, there is substantial interest in electronic banking that enables people to access their account information from home. This includes traditional bank checking and savings accounts, and real-time information on stocks, futures, and currency trading accounts. Such applications can eliminate many of the agencies involved in account management by enabling consumers to electronically manage their accounts. These applications can also generate a lot of revenue to help pay for the investments in broadband networks. Although these applications do not require high-bandwidth channels for communications, they do require secure point-to-point communication channels. Therefore, the Internet in its current form is not suitable for electronic banking applications (as described later).
Point-to-point channels can be easily allocated in a switched fiber coax network, but are more difficult in a broadcast network. Broadcast networks have to use some other medium (e.g., telephone network) as the two-way signaling network. Alternatively, one can create virtual point-to-point broadband channels between a server and a terminal by multiplexing all such point-to-point channels onto one high-bandwidth channel and enabling terminals to access and decode information meant for them. More secure and expensive solutions can be implemented by encrypting the entire virtual channel for a terminal using a key, and performing on-the-fly decryption of the virtual channel on the terminal using its secure key.
Downloaded Applications
Interactive applications can be provided in two ways. The centralized method is to perform all processing at a central server and transmit responses to the end user's terminal. The distributed method has processing being performed in both the terminal and the server, and follows the familiar client-server paradigm. The latter approach requires the client part of the application to be downloaded to the user's terminal [8]. Such applications are called "downloaded applications."
Being able to execute software locally in the network terminal reduces server and network costs, and greatly expands the range of services available to the end user. The decreased cost and increased scope of services are due to the reduced response time of local software compared to responses generated remotely, which are limited by network and server latencies. Trying to overcome these latencies by increasing server and network performance may be more costly than a slight increase in terminal flexibility.
The MPEG-2 standard for compressed video and audio also provides the Digital Storage Media Command & Control (DSM-CC) standard (ISO/IEC 13818/6) [14]. This standard describes protocols to be used for client/server communication, including protocols for downloading code and data from a server to a client.
Application Download in the Dover Deployment
Network terminals in the Dover Township deployment have rudimentary control software in read-only memory that provides a television-like interface to the end user. Additionally, this software gives the end user the capability to download more full-featured applications. Since an interpreter is not available in the read-only memory, one must first be downloaded if interpreted code (e.g., Java [9], MHEG [10]) is to be executed.
Data for download is broadcast repetitively from data carousels at the service provider's site. When the end user requests an application download, the terminal connects to the service provider. The service provider transmits information about the broadcast stream carrying the download data. Using this information, the terminal asks the network to switch the broadcast stream to its network address. It captures the data being broadcast, reassembles the original image of the software, and then executes it. The downloaded software then proceeds to interact with the end user and a server on the network.
The network terminal allows applications to communicate back to a server on the network. Thus, an application can request specific data or applets to be transmitted for download. Portions of the application can be downloaded and then discarded as necessary. This allows a large application to execute within the memory limits of the network terminal.
The capability of a home network terminal to download applications can also be used to download new versions of system software for the terminal. This capability of the network terminal is being used extensively in the Dover deployment to distribute upgrades to the resident operating system without having to send maintenance personnel to consumers' homes. The operating system upgrade mechanism has two steps, downloading and verifying a new binary image of the operating system, and overwriting the old operating system with the new version. Downloading a new binary image of the operating system is very similar to downloading applications. Overwriting the resident operating system stored in NVRAM is a critical operation because failure can cause the terminal to be in an inconsistent state.
Interactive Video Networks and the Internet
There are very few switched broadband networks in existence today that are designed primarily to carry real-time audio and video. The network in Dover Township, New Jersey is one of the largest commercially deployed switched interactive digital video networks. All other major U.S. telecommunication companies have stopped or scaled back plans for such networks. In the meantime, the Internet is expanding at an exponential rate and providing e-mail, chat, and other interactive services to millions of people worldwide.
There are currently efforts underway to merge the services provided by broadband video networks and the Internet. The Dover network has many of the basic capabilities needed to support Internet applications. The bandwidth between the network terminals and servers is adequate in both directions for Internet applications. Internet applications require TCP/IP protocols to communicate between servers and terminals, whereas the Dover network uses ATM for broadband communications and a proprietary protocol over X.25 for two-way communications. Thus, supporting Internet applications requires a Transmission Control Protocol/Internet Protocol (TCP/IP) protocol stack to be built on top of the proprietary two-way protocol. It also requires building all the software needed to run a browser and a Java runtime environment.
DAVIC
The Digital Audio Video Council (DAVIC), an international organization specifying the interfaces and services for broadband video networks [11], has specified many variants of broadband networks, including switched fiber coax, satellite, and xDSL. It has also specified physical interfaces, data flows, distribution formats, and protocols between content servers, network managers, and home equipment.
DAVIC is now changing directions and focusing on the convergence of Internet and broadband video services by repositioning broadband networks as a transport medium for future Internet services requiring quality of service (QoS) guarantees. It has specified Internet access to consumers on DAVIC video networks as a step in the new direction.
Security Issues: Internet vs. Broadband Networks
Digital video information over the Internet has to address the question of security. The Internet provides very little protection to information carried on it. Digital video signals would need to be encrypted by the sender and decrypted by the receiver. While this is no less secure than the scrambling mechanisms used in satellite or MMDS networks, the real issues are authenticating the customer and ensuring that sensitive buying information provided by the consumer does not fall into the wrong hands.
In applications such as home shopping and electronic banking, a private network built by the local telephone or cable operator provides a secure channel to the service provider. The privacy and authenticity of information on the private network can be ensured by the network operator, who is often the service provider as well. On the Internet, the service provider can only control one end point, and has limited choices when it comes to security. Public-key encryption schemes such as Rivest, Shamir, Adleman (RSA) [12] can be used for authentication in this environment, but require that the service provider store the public keys of subscribing customers. Depending on the service architecture, this may or may not be feasible. Most services over the Internet today accept information such as credit-card numbers in an un-encrypted message. The security risks of such a mechanism are obvious.
The advent of mobile code such as Java [9] and limbo [13] show some promise that information can be transmitted more securely over the Internet. Using these mechanisms, the service provider can download a fragment of code on the customer machine, which can encrypt sensitive information in any manner desired by the service provider and send it to him in a secure manner. These approaches, in conjunction with strong encryption schemes, may solve the security problems of the current Internet.
Conclusions
Various broadband networks have been trialed. These have shown that home users are unwilling to pay high prices for video on demand, necessitated by the huge infrastructure demands of that application. However, value-added end-to-end applications and services for interactive video networks have the potential for substantial growth in the future. Such services include videoconferencing, home shopping, virtual travel agencies, electronic banking services, cheaper long distance calling, and so on. Today, users are willing to pay for such services over the Internet [5] even though the Internet is not capable of providing real-time multimedia services. The combination of broadband networks and multimedia home terminals can unleash a new generation of value-added applications.
The rapid growth of the Internet and the Web, in spite of the limited bandwidth, is testimony to the fact that new, easy-to-use applications and services drive the acceptance of technology. Broadband video networks need to support the Internet and other new end-to-end applications that make effective use of the available bandwidth and are clearly superior to existing services. It should also be understood that consumers need many forms of interactivity, and unidirectional broadband networks with very low-bandwidth two-way communication may not be acceptable for many applications (e.g., video conferencing).
In the future we are likely to see an amalgamation of interactive video networks and the Internet. It will be a requirement for all networks to support Internet services because of the amount of information available on the Internet. Broadband networks will help provide transport services for real-time data and security for various Internet applications.
References
[1] C. Blank, "The FSN Challenge: Large-Scale Interactive Television," IEEE Computer, vol. 28, May 1995, pp. 9–12.
[2] S. Ramanathan and P. V. Rangan, "Architectures for Personalized Multimedia," IEEE Multimedia, vol. 1, Spring 1994, pp. 37–46.
[3] D. Large, "Creating a Network for Interactivity," IEEE Spectrum, vol. 32, April 1995, Special Issue on Digital Television, pp. 58–63.
[4] S. Dixit, "A Look at the Video Dial-Tone Network," IEEE Spectrum, vol. 32, Apr. 1995, Special Issue on Digital Television, pp. 64–65.
[5] L. Press, "The Internet and Interactive Television," Commun. ACM, vol. 36, Dec. 1993, pp. 19–23.
[6] CCITT, Rec. X.25, "Public Packet Switched Networks".
[7] J. W. Darr, "Dover Township Video Dialtone Network Architecture," Bell Atlantic Video Information Provider Workshop, 1995.
[8] L. Chariglione, "MPEG: A Technological Basis for Multimedia Applications," IEEE Multimedia, vol. 2, Spring 1995, pp. 85–89.
[9] K. Arnold and J. Gosling, The Java Programming Language, Reading, MA: Addision Wesley, 1996, ISBN 0201634554.
[10] T. Meyer-Boudnik and W. Effelsberg, "MHEG Explained," IEEE Multimedia, vol. 2, Spring 1995, pp. 26–38.
[11] "Digital Audio Video Council (DAVIC) specification 1.1", Sept. 1996.
[12] A. N. S. Instricture, "American National Standard x9.31-1992: Public Key Cryptography Using Reversible Algorithms for the Financial Services Industry, Part 1: The RSA Signature Algorithm," Mar. 1993.
[13] B. Kernighan, "A Descent into Limbo", July 1996.
[14] "Coding for Motion Pictures and Associated Video MPEG-2", ISO/IEC/SC29/WG11, 1994.
Biography
Kamlesh Rath [M] is a Senior Member of Research Staff at Philips Research, Briarcliff Manor, New York. He received his B.E. in computer science and engineering from Jadavpur University, Calcutta, India in 1987, and a Ph.D. in computer science from Indiana University, Bloomington, in 1995. His research interests are in the areas of multimedia operating systems, digital TV systems, reconfigurable architectures, and hardware-software co-design.
Don H. Wanigasekara-Mohotti [M] received a B.Sc. in electrical engineering from Columbia University, New York in 1991, and a M.Sc. in computer science from Polytechnic University, New York, in 1996. He currently works in the Software Systems and Architecture Department at Philips Research, New York. His interests include networked multimedia systems, user interfaces, and graphics modeling and rendering systems.
Roli G. Wendorf is a Senior Member of Research Staff at Philips Research, Briarcliff Manor, New York. Since 1993, she has been involved in projects related to software for digital interactive TV and HDTV receivers. She has led software process improvement (SPI) activities in Philips' research and development environment for two years. From 1987 to 1993, Roli worked in the research and development of production control systems for automated discrete manufacturing. She served as the chief architect of these systems for three years. In addition to SPI, her current interests are in the areas of software engineering, digital multimedia, real-time systems, and distributed systems. Roli has an M.S. in computer science (1986) from Carnegie-Mellon University, and M.S. (1979) and B.S. (1976) degrees in electrical engineering from the Indian Institute of Technology.
Dinesh C. Verma [M] received a B.Tech in computer science from Indian Institute of Technology, Kanpur, in 1987, and a Ph.D. from the University of California, Berkeley in 1991. After his doctorate, he worked at the IBM T. J. Watson Research Center, Yorktown Heights, New York, in the area of high-speed networks, and then joined the Software Systems and Architecture Department at Philips Research in 1995. His research interests include computer networks, quality of service issues, multimedia operating systems, and distributed computing.